CN110781901B - Instrument ghost character recognition method based on BP neural network prediction threshold - Google Patents
Instrument ghost character recognition method based on BP neural network prediction threshold Download PDFInfo
- Publication number
- CN110781901B CN110781901B CN201911038311.6A CN201911038311A CN110781901B CN 110781901 B CN110781901 B CN 110781901B CN 201911038311 A CN201911038311 A CN 201911038311A CN 110781901 B CN110781901 B CN 110781901B
- Authority
- CN
- China
- Prior art keywords
- image
- neural network
- ghost
- instrument
- gray level
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Active
Links
- 238000000034 method Methods 0.000 title claims abstract description 36
- 238000013528 artificial neural network Methods 0.000 title claims abstract description 33
- 238000012937 correction Methods 0.000 claims abstract description 5
- PXFBZOLANLWPMH-UHFFFAOYSA-N 16-Epiaffinine Natural products C1C(C2=CC=CC=C2N2)=C2C(=O)CC2C(=CC)CN(C)C1C2CO PXFBZOLANLWPMH-UHFFFAOYSA-N 0.000 claims abstract description 4
- 230000009466 transformation Effects 0.000 claims abstract description 4
- 238000012549 training Methods 0.000 claims description 23
- 230000006870 function Effects 0.000 claims description 11
- 238000010606 normalization Methods 0.000 claims description 3
- 238000005457 optimization Methods 0.000 claims description 3
- 238000012546 transfer Methods 0.000 claims description 3
- 238000000926 separation method Methods 0.000 claims 1
- 230000011218 segmentation Effects 0.000 abstract description 2
- 238000010586 diagram Methods 0.000 description 3
- 230000004913 activation Effects 0.000 description 2
- 238000013527 convolutional neural network Methods 0.000 description 2
- 230000008034 disappearance Effects 0.000 description 2
- 230000000694 effects Effects 0.000 description 2
- 238000005286 illumination Methods 0.000 description 2
- 238000004891 communication Methods 0.000 description 1
- 230000007547 defect Effects 0.000 description 1
- 230000001419 dependent effect Effects 0.000 description 1
- 238000005516 engineering process Methods 0.000 description 1
- 238000005259 measurement Methods 0.000 description 1
- 238000011176 pooling Methods 0.000 description 1
- 230000008569 process Effects 0.000 description 1
- 238000012545 processing Methods 0.000 description 1
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V30/00—Character recognition; Recognising digital ink; Document-oriented image-based pattern recognition
- G06V30/10—Character recognition
- G06V30/14—Image acquisition
- G06V30/148—Segmentation of character regions
- G06V30/153—Segmentation of character regions using recognition of characters or words
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/04—Architecture, e.g. interconnection topology
- G06N3/048—Activation functions
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/08—Learning methods
- G06N3/084—Backpropagation, e.g. using gradient descent
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T3/00—Geometric image transformations in the plane of the image
- G06T3/02—Affine transformations
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T5/00—Image enhancement or restoration
- G06T5/70—Denoising; Smoothing
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T7/00—Image analysis
- G06T7/10—Segmentation; Edge detection
- G06T7/11—Region-based segmentation
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T7/00—Image analysis
- G06T7/10—Segmentation; Edge detection
- G06T7/136—Segmentation; Edge detection involving thresholding
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V2201/00—Indexing scheme relating to image or video recognition or understanding
- G06V2201/02—Recognising information on displays, dials, clocks
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Biophysics (AREA)
- Data Mining & Analysis (AREA)
- Life Sciences & Earth Sciences (AREA)
- Artificial Intelligence (AREA)
- Biomedical Technology (AREA)
- Software Systems (AREA)
- Computational Linguistics (AREA)
- Health & Medical Sciences (AREA)
- Evolutionary Computation (AREA)
- General Health & Medical Sciences (AREA)
- Molecular Biology (AREA)
- Computing Systems (AREA)
- General Engineering & Computer Science (AREA)
- Mathematical Physics (AREA)
- Multimedia (AREA)
- Character Input (AREA)
Abstract
The invention relates to the field of image recognition, and discloses an instrument ghost character recognition method based on a BP neural network prediction threshold, which comprises the following steps: obtaining a color image of a digital display instrument, carrying out graying treatment to obtain a gray image, calculating gray level distribution statistics of the gray image, inputting the gray level distribution statistics into a BP neural network, predicting an ideal binary global threshold, carrying out binarization on the gray image to obtain a binary image for eliminating ghost, carrying out small connected domain removal treatment on the binary image, creating a minimum circumscribed rectangle of the binary image, calculating an inclination angle, realizing inclination correction through affine transformation, dividing the corrected binary image into single character images through a projection segmentation method, normalizing the character image size to be 32 multiplied by 32, inputting the character images into a LeNet-5 model for identification, and obtaining an identification result of the digital display instrument characters. The instrument ghost character recognition method based on the BP neural network prediction threshold effectively overcomes the influence of ghost, and is high in recognition rate and high in recognition speed.
Description
Technical Field
The invention relates to the field of image recognition, in particular to an instrument ghost character recognition method based on a BP neural network prediction threshold.
Background
The accuracy of the electronic measuring instrument is critical for measurement, and the accuracy of the electronic measuring instrument needs to be detected regularly. The traditional digital display type instrument lacks a communication interface, cannot directly acquire the measured value of the instrument, can only acquire images of the standard meter and the measured meter in the rapid boosting and reducing processes respectively through double cameras, automatically identifies the reading of the instrument, and finally compares the results of the standard meter and the measured meter to judge the accuracy of the measured meter. However, when the number of meters changes, serious ghosts occur in part of the meters, and the ghosts increase the difficulty of binarizing the image.
The commonly used binarization methods are mainly divided into two main categories: global threshold algorithms and local threshold algorithms. The global threshold algorithm mainly comprises an law method (Otsu), a maximum entropy method, an iteration method and the like, and is mainly applicable to pictures with uniform illumination and obvious double peaks of gray histograms. The local threshold algorithm mainly comprises a Sauvla algorithm, a Niblack algorithm, a Bernsen algorithm and the like, and is mainly applicable to pictures with uneven illumination. All the above algorithms can only separate the background of the meter image and cannot separate the digital ghost.
In addition, algorithms such as SVM, BP neural network, template matching and KNN are widely applied to instrument character recognition, and good recognition effects are obtained. The recognition effect of the above algorithm is largely dependent on the distinguishability of the extracted features. The convolutional neural network can automatically extract proper features for classification, overcomes the difficulty of feature dependence of the traditional algorithm, and is suitable for recognition of instrument characters. The LeNet-5 is a classical convolutional neural network suitable for handwriting character recognition, and mainly comprises an input layer, an output layer, two convolutional layers, two pooling layers and three full-connection layers, and has guiding significance for recognition of instrument characters.
Disclosure of Invention
Aiming at the defects of the technology, the invention provides the instrument ghost character recognition method based on the BP neural network prediction threshold, which effectively overcomes the influence of ghost, and has high recognition rate and high recognition speed.
In order to achieve the above purpose, the instrument ghost character recognition method based on the BP neural network prediction threshold, which is designed by the invention, comprises the following steps:
a) Acquiring a color image of the digital display instrument through a camera;
b) Graying the color image obtained in the step A) to obtain a gray image;
c) Calculating gray level distribution statistics of the gray level image obtained in the step B), inputting the gray level distribution statistics into a BP neural network, predicting an ideal binarization global threshold value, and binarizing the gray level image to obtain a binary image for eliminating ghost images;
d) The small connected domain is removed from the binary image obtained in the step C), so that the influence of noise is reduced;
e) Creating a minimum circumscribed rectangle of the binary image, calculating an inclination angle, and realizing inclination correction through affine transformation;
f) Dividing the corrected binary image into single character images by adopting a projection division method;
g) Normalizing the character image size obtained in the step F) to 32×32;
h) And inputting the character image with normalized size into a LeNet-5 model for recognition, and obtaining a recognition result of the digital display instrument characters.
Preferably, in the step C), the step a) is repeated to obtain a training sample image, gray scale is performed on the training sample image, gray scale distribution statistics are extracted, an ideal binary global threshold is obtained, the gray scale distribution statistics are used as input, a corresponding ideal binary global threshold is used as output, and the BP neural network is trained.
Preferably, the selection of the ideal binarized global threshold is required to meet the criteria that the images after binarization do not contain ghost images and the numbers after the ghost images are separated are complete, the maximum value and the minimum value of the reasonable global threshold of each training sample image are manually selected, and the average value of the reasonable global thresholds is calculated to be used as the ideal binarized global threshold of the training sample image.
Preferably, in the step C), the BP neural network includes 2 hidden layers, the number of nodes of each hidden layer is 25, the number of input nodes corresponds to the gray level distribution statistic dimension, 256 is an output node number is 1, the node transfer functions of the hidden layers and the output layer of the BP neural network are Tansig, and the training function is Traingdm.
Preferably, the LeNet-5 model in the step H) adopts a ReLU to replace the sigmoid function, so that the problems of gradient disappearance and poor generalization capability of the sigmoid activation function are avoided, and the convergence speed is higher.
Preferably, the LeNet-5 model in the step H) updates the weight and bias by adopting an RMSprop optimization algorithm, so that the problems of difficult learning rate selection and local minimum sinking of the gradient descent algorithm are avoided, and the convergence speed is higher.
Preferably, in the step H), the training of the LeNet-5 model takes a single character image obtained by dividing a training sample and normalizing the size as input, and takes a corresponding label manually produced as output.
Preferably, in the step G), the normalization method is scaled and centered in equal proportion, and black is filled around.
Compared with the prior art, the invention has the following advantages: the method realizes the identification of the ghost characters of the instrument, effectively overcomes the influence of ghost, and has high identification rate and high identification speed.
Drawings
FIG. 1 is a flow chart of the instrument ghost character recognition method based on BP neural network prediction threshold;
FIG. 2 is a color image acquired by a camera;
FIG. 3 is a gray scale of FIG. 2;
FIG. 4 is a binary diagram of FIG. 3;
FIG. 5 is a flow chart of tilt correction;
FIG. 6 is a flow chart of projection segmentation;
FIG. 7 is a schematic diagram of normalized character images;
FIG. 8 is a schematic diagram of the LeNet-5 model.
Detailed Description
The invention will now be described in further detail with reference to the drawings and to specific examples.
An instrument ghost character recognition method based on BP neural network prediction threshold, as shown in figure 1, comprises the following steps:
a) As shown in fig. 2, a color image of the digital display instrument is acquired through a camera;
b) As shown in fig. 3, the color image obtained in step a) is subjected to graying processing to obtain a gray image;
c) Calculating gray level distribution statistics of the gray level image obtained in the step B), inputting the gray level distribution statistics into a BP neural network, predicting an ideal binarization global threshold value, and binarizing the gray level image, as shown in fig. 4, to obtain a binary image for eliminating double images;
d) The small connected domain is removed from the binary image obtained in the step C), so that the influence of noise is reduced;
e) Creating a minimum circumscribed rectangle of the binary image, calculating an inclination angle, and realizing inclination correction through affine transformation, as shown in fig. 5;
f) As shown in fig. 6, the corrected binary image is divided into single character images by adopting a projection division method;
g) As shown in fig. 7, the character image size obtained in step F) is normalized to 32×32, the normalization method is scaled equally and centered, and black is filled around;
h) And inputting the character image with normalized size into a LeNet-5 model for recognition, and obtaining a recognition result of the digital display instrument characters as shown in fig. 8.
In this embodiment, in step C), repeating step a), obtaining a training sample image, graying the training sample image, extracting gray level distribution statistics therein, obtaining an ideal binary global threshold, taking the gray level distribution statistics as input, and taking a corresponding ideal binary global threshold as output, and training the BP neural network, where the selection of the ideal binary global threshold is required to meet the criteria that the binarized image does not contain ghost and has complete digits after the ghost is separated, manually selecting the maximum value and the minimum value of the reasonable global threshold of each training sample image, and obtaining the average value of the reasonable global thresholds as the ideal binary global threshold of the training sample image. In this embodiment, the trained BP neural network includes 2 hidden layers, the number of nodes in each hidden layer is 25, the number of input nodes corresponds to the gray level distribution statistic dimension, 256, the number of output nodes is 1, the node transfer functions of the hidden layers and the output layers of the BP neural network are Tansig, and the training function is Traingdm. Inputting the gray level distribution statistic of the gray level image obtained in the step B) into the trained BP neural network, and predicting an ideal binarization global threshold.
In the embodiment, the LeNet-5 model in the step H) adopts a ReLU to replace a sigmoid function, so that the problems that the sigmoid activation function has gradient disappearance and poor generalization capability are avoided, the weight and bias are updated by adopting an RMSprop optimization algorithm, the problems that the learning rate of a gradient descent algorithm is difficult to select and the gradient descent algorithm falls into a local minimum value are avoided, and the convergence speed is higher. In step H), training of the LeNet-5 model takes a single character image which is obtained by dividing a training sample and normalizing the size as input, and takes a corresponding label which is manually manufactured as output.
According to the instrument ghost character recognition method based on the BP neural network prediction threshold, instrument images acquired by the camera are recognized into instrument numbers through internal recognition, so that the recognition of instrument ghost characters is realized, the influence of ghost is effectively overcome, the recognition rate is high, and the recognition speed is high.
Claims (8)
1. The instrument ghost character recognition method based on the BP neural network prediction threshold is characterized by comprising the following steps of: the method comprises the following steps:
a) Acquiring a color image of the digital display instrument through a camera;
b) Graying the color image obtained in the step A) to obtain a gray image;
c) Calculating gray level distribution statistics of the gray level image obtained in the step B), inputting the gray level distribution statistics into a BP neural network, predicting an ideal binarization global threshold value, and binarizing the gray level image to obtain a binary image for eliminating ghost images;
d) Performing small connected domain removal treatment on the binary image obtained in the step C);
e) Creating a minimum circumscribed rectangle of the binary image, calculating an inclination angle, and realizing inclination correction through affine transformation;
f) Dividing the corrected binary image into single character images by adopting a projection division method;
g) Normalizing the character image size obtained in the step F) to 32×32;
h) And inputting the character image with normalized size into a LeNet-5 model for recognition, and obtaining a recognition result of the digital display instrument characters.
2. The method for identifying the ghost characters of the instrument based on the BP neural network prediction threshold according to claim 1, wherein the method comprises the following steps: in the step C), repeating the step a), obtaining a training sample image, graying the training sample image, extracting gray level distribution statistics therein, obtaining an ideal binarization global threshold, taking the gray level distribution statistics as input, and taking a corresponding ideal binarization global threshold as output, and training the BP neural network.
3. The method for identifying the ghost characters of the instrument based on the BP neural network prediction threshold according to claim 2, wherein the method comprises the following steps: the ideal binarization global threshold is selected to meet the standard that the images after binarization do not contain double images and the numbers after separation of double images are complete, the maximum value and the minimum value of the reasonable global threshold of each training sample image are manually selected, and the average value of the reasonable global threshold is calculated to be used as the ideal binarization global threshold of the training sample image.
4. The method for identifying the ghost characters of the instrument based on the BP neural network prediction threshold according to claim 3, wherein the method comprises the following steps: in the step C), the BP neural network includes 2 hidden layers, the number of nodes in each hidden layer is 25, the number of input nodes corresponds to the gray level distribution statistic dimension, 256 is output nodes is 1, the node transfer functions of the hidden layers and the output layers of the BP neural network are Tansig, and the training function is Traingdm.
5. The method for identifying the ghost characters of the instrument based on the BP neural network prediction threshold according to claim 1, wherein the method comprises the following steps: the LeNet-5 model in step H) uses a ReLU to replace the sigmoid function.
6. The method for identifying the ghost characters of the instrument based on the BP neural network prediction threshold according to claim 1, wherein the method comprises the following steps: the LeNet-5 model in the step H) adopts an RMSprop optimization algorithm to update the weight and bias.
7. The method for identifying the ghost characters of the instrument based on the BP neural network prediction threshold according to claim 1, wherein the method comprises the following steps: in the step H), the training of the LeNet-5 model takes a single character image which is obtained by dividing a training sample and normalizing the size as input, and a corresponding label which is manually manufactured as output.
8. The method for identifying the ghost characters of the instrument based on the BP neural network prediction threshold according to claim 1, wherein the method comprises the following steps: in the step G), the normalization method is scaled and centered in equal proportion, and black is filled around.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201911038311.6A CN110781901B (en) | 2019-10-29 | 2019-10-29 | Instrument ghost character recognition method based on BP neural network prediction threshold |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201911038311.6A CN110781901B (en) | 2019-10-29 | 2019-10-29 | Instrument ghost character recognition method based on BP neural network prediction threshold |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN110781901A CN110781901A (en) | 2020-02-11 |
| CN110781901B true CN110781901B (en) | 2023-04-28 |
Family
ID=69387361
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201911038311.6A Active CN110781901B (en) | 2019-10-29 | 2019-10-29 | Instrument ghost character recognition method based on BP neural network prediction threshold |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN110781901B (en) |
Families Citing this family (8)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN111428069A (en) * | 2020-03-11 | 2020-07-17 | 中交第二航务工程局有限公司 | Construction data acquisition method for slot milling machine |
| CN111383199B (en) * | 2020-03-23 | 2023-05-26 | 如你所视(北京)科技有限公司 | Image processing method, device, computer readable storage medium and electronic equipment |
| CN111935464B (en) * | 2020-10-19 | 2021-03-12 | 太仓市昊博技术服务有限公司 | Brightness recognition instrument monitoring device and method |
| CN112487981A (en) * | 2020-11-30 | 2021-03-12 | 哈尔滨工程大学 | MA-YOLO dynamic gesture rapid recognition method based on two-way segmentation |
| CN113221737B (en) * | 2021-05-11 | 2023-09-05 | 杭州海康威视数字技术股份有限公司 | Material information determining method, device, equipment and storage medium |
| CN113343969A (en) * | 2021-06-21 | 2021-09-03 | 重庆科技学院 | GBS-YOLOv5 s-based pointer instrument intelligent digital display method |
| CN113888536B (en) * | 2021-12-01 | 2022-03-18 | 武汉市利友印务有限公司 | Printed matter double image detection method and system based on computer vision |
| CN116721042A (en) * | 2023-08-10 | 2023-09-08 | 广东石油化工学院 | Multi-threshold binarization-based image tilt correction method |
Citations (10)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101021905A (en) * | 2006-02-15 | 2007-08-22 | 中国科学院自动化研究所 | File image binaryzation method |
| CN101567041A (en) * | 2009-05-25 | 2009-10-28 | 公安部交通管理科学研究所 | Method for recognizing characters of number plate images of motor vehicles based on trimetric projection |
| CN104200211A (en) * | 2014-09-03 | 2014-12-10 | 腾讯科技（深圳）有限公司 | Image binaryzation method and device |
| CN107292873A (en) * | 2017-06-29 | 2017-10-24 | 西安工程大学 | A kind of close degree detecting method of porcelain insulator ash based on color character |
| JP2017207910A (en) * | 2016-05-18 | 2017-11-24 | 株式会社東芝 | Character recognition device, character recognition program and character recognition method |
| CN107945200A (en) * | 2017-12-14 | 2018-04-20 | 中南大学 | Image binaryzation dividing method |
| CN109035274A (en) * | 2018-08-23 | 2018-12-18 | 湖北工业大学 | File and picture binary coding method based on background estimating Yu U-shaped convolutional neural networks |
| CN109086762A (en) * | 2018-07-20 | 2018-12-25 | 上海工程技术大学 | A kind of fuzzy license plate identification method based on more mesh cameras |
| CN109740603A (en) * | 2019-01-21 | 2019-05-10 | 闽江学院 | Based on the vehicle character identifying method under CNN convolutional neural networks |
| CN110210477A (en) * | 2019-05-24 | 2019-09-06 | 四川阿泰因机器人智能装备有限公司 | A kind of digital instrument Recognition of Reading method |
Family Cites Families (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US8351699B2 (en) * | 2008-12-18 | 2013-01-08 | Accusoft Corporation | Methods and apparatus for auto image binarization |
-
2019
- 2019-10-29 CN CN201911038311.6A patent/CN110781901B/en active Active
Patent Citations (10)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101021905A (en) * | 2006-02-15 | 2007-08-22 | 中国科学院自动化研究所 | File image binaryzation method |
| CN101567041A (en) * | 2009-05-25 | 2009-10-28 | 公安部交通管理科学研究所 | Method for recognizing characters of number plate images of motor vehicles based on trimetric projection |
| CN104200211A (en) * | 2014-09-03 | 2014-12-10 | 腾讯科技（深圳）有限公司 | Image binaryzation method and device |
| JP2017207910A (en) * | 2016-05-18 | 2017-11-24 | 株式会社東芝 | Character recognition device, character recognition program and character recognition method |
| CN107292873A (en) * | 2017-06-29 | 2017-10-24 | 西安工程大学 | A kind of close degree detecting method of porcelain insulator ash based on color character |
| CN107945200A (en) * | 2017-12-14 | 2018-04-20 | 中南大学 | Image binaryzation dividing method |
| CN109086762A (en) * | 2018-07-20 | 2018-12-25 | 上海工程技术大学 | A kind of fuzzy license plate identification method based on more mesh cameras |
| CN109035274A (en) * | 2018-08-23 | 2018-12-18 | 湖北工业大学 | File and picture binary coding method based on background estimating Yu U-shaped convolutional neural networks |
| CN109740603A (en) * | 2019-01-21 | 2019-05-10 | 闽江学院 | Based on the vehicle character identifying method under CNN convolutional neural networks |
| CN110210477A (en) * | 2019-05-24 | 2019-09-06 | 四川阿泰因机器人智能装备有限公司 | A kind of digital instrument Recognition of Reading method |
Non-Patent Citations (2)
| Title |
|---|
| Zhiwen Wang.Binarization Algorithm of Passport Image Based on Global Iterative Threshold and Local Analysis.《2009 International Conference on Information Management, Innovation Management and Industrial Engineering》.2009,全文. * |
| 廖原.基于卷积网络的车牌二值化算法.《信息与电脑(理论版)》.2018,全文. * |
Also Published As
| Publication number | Publication date |
|---|---|
| CN110781901A (en) | 2020-02-11 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN110781901B (en) | Instrument ghost character recognition method based on BP neural network prediction threshold | |
| CN108122236B (en) | Iterative fundus image blood vessel segmentation method based on distance modulation loss | |
| CN107545239B (en) | Fake plate detection method based on license plate recognition and vehicle characteristic matching | |
| CN107103317A (en) | Fuzzy license plate image recognition algorithm based on image co-registration and blind deconvolution | |
| CN111539330B (en) | Transformer substation digital display instrument identification method based on double-SVM multi-classifier | |
| CN112270317B (en) | Reading identification method of traditional digital water meter based on deep learning and frame difference method | |
| CN106529532A (en) | License plate identification system based on integral feature channels and gray projection | |
| CN111652213A (en) | Ship water gauge reading identification method based on deep learning | |
| CN113888536B (en) | Printed matter double image detection method and system based on computer vision | |
| CN113379833B (en) | Image visible watermark positioning and segmenting method based on neural network | |
| CN109635726B (en) | Landslide identification method based on combination of symmetric deep network and multi-scale pooling | |
| CN110689060B (en) | Heterogeneous image matching method based on aggregation feature difference learning network | |
| CN115994907B (en) | Intelligent processing system and method for comprehensive information of food detection mechanism | |
| CN113221956B (en) | Target identification method and device based on improved multi-scale depth model | |
| CN116188880B (en) | Cultivated land classification method and system based on remote sensing image and fuzzy recognition | |
| CN111553873A (en) | Automatic brain neuron detection method based on multi-scale convolutional neural network | |
| CN118038515B (en) | Face recognition method | |
| CN117173158A (en) | Intelligent detection method and system for quality of precise connector | |
| CN115100406B (en) | Weight information entropy fuzzy C-means clustering method based on superpixel processing | |
| CN115033721A (en) | Image retrieval method based on big data | |
| CN110992301A (en) | Gas contour identification method | |
| CN113989256A (en) | Detection model optimization method, detection method and detection device for remote sensing image building | |
| CN113723410B (en) | Digital identification method and device for nixie tube | |
| CN116958126A (en) | Method and system for detecting injection molding of mold based on computer vision | |
| CN108710915B (en) | Multi-feature fusion gastroscope image processing method based on multi-kernel learning |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant | ||
| EE01 | Entry into force of recordation of patent licensing contract | ||
| EE01 | Entry into force of recordation of patent licensing contract |
Application publication date: 20200211 Assignee: Nanjing Ruishijie Electronic Technology Co.,Ltd. Assignor: HUBEI University OF TECHNOLOGY Contract record no.: X2023980049595 Denomination of invention: Instrument Ghost Character Recognition Method Based on BP Neural Network Predicting Thresholds Granted publication date: 20230428 License type: Exclusive License Record date: 20231205 |